Fix convert_int_to_float in compact_incomplete V1 library API method #2235
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
vasil-pashov
requested review from
alexowens90,
willdealtry and
poodlewars
as code owners
vasil-pashov
force-pushed
the
vasil.pashov/fix-convert-int-to-float
branch
from
4fa6bb9 to
23af9d8
Compare
IvoDD
approved these changes
Mar 17, 2025
There was a problem hiding this comment.
Your tests pretty much cover all cases with convert_int_to_float=True.
What do you think about also adding tests that we get schema exseptions if convert_int_to_float=False? (e.g. to ensure we don't accidentally pass convert_int_to_float=True somewhere deep in C++ where we shouldn't). It is possible we already have a bunch of these tests somewhere else, haven't looked through them
alexowens90
reviewed
Mar 17, 2025
| auto new_descriptor = has_valid_common_type(existing_type_desc, type_desc); | ||
| if(new_descriptor) { | ||
| merged_fields_map[field.name()] = *new_descriptor; | ||
| if (convert_int_to_float && is_integer_type(existing_type_desc.data_type()) && is_integer_type(type_desc.data_type())) { |
There was a problem hiding this comment.
What if the existing type is float64? Or float32?
alexowens90
reviewed
Mar 17, 2025
| assert_frame_equal(expected, lib.read(sym).data, check_dtype=True) | ||
|
|
||
| @pytest.mark.parametrize("dtype1", [np.int32, np.uint16, np.int8, np.int64, np.uint64, np.float64]) | ||
| @pytest.mark.parametrize("dtype2", [np.int32, np.uint16, np.int8, np.int64, np.uint64, np.float64]) |
There was a problem hiding this comment.
Include float32 in the tested dtypes?
alexowens90
reviewed
Mar 17, 2025
| expected = pd.DataFrame({"a": np.arange(1, 7, dtype=np.double)}, index=pd.date_range(pd.Timestamp(0), periods=6, freq="ns")) | ||
| assert_frame_equal(expected, lib.read(sym).data, check_dtype=True) | ||
|
|
||
| @pytest.mark.parametrize("dtype2", [np.int32, np.uint16, np.int8, np.int64, np.uint64, np.float64]) |
vasil-pashov
force-pushed
the
vasil.pashov/fix-convert-int-to-float
branch
from
37adac2 to
3d55226
Compare
alexowens90
approved these changes
Mar 18, 2025
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Reference Issues/PRs
What does this implement or fix?
This fixes the behavior of
convert_int_to_floatexposed via the V1 Library API'scompact_incompletemethod.Convert int to float will change the type of all integer columns (both signed and unsigned) to float64. In contrast to dynamic schema conversion this will static cast the values of all segments to float64 prior writing. This functionality was broken when we made the stricter type checks for staged segments.
Additional bugfix was added for dynamic schema. Dynamic schema has a different code path as it calls merge descriptors when it collects all incomplete segments thus the option was not properly applied even in arcticc. Dynamic schema used to throw exception when unsigned and signed int were mixed together (in the same column) but this should not matter as they are both going to be cast to double.
There are two main points:
Any other comments?
Checklist
Checklist for code changes...